Replication Protein A2 Phosphorylation after DNA Damage by the Coordinated Action of Ataxia Telangiectasia-Mutated and DNA-dependent Protein Kinase ¹

RPA³ (also known as human ssDNA binding protein) is a trimeric protein complex involved in many cellular processes including DNA replication initiation and elongation, DNA repair, and recombination (1, 2). Human RPA is a heterotrimer composed of M_r 70,000 (RPA1), M_r 29,000 (RPA2), and M_r 14,000 (RPA3) subunits (3, 4). RPA2 is a phosphoprotein that becomes differentially phosphorylated throughout the cell cycle. Phosphorylation of RPA2 is first observed at the G₁-S transition and is maintained through late mitosis (5, 6). In vitro, phosphorylation of RPA2 occurs during SV40 DNA replication, and binding of RPA to ssDNA stimulates this modification (7, 8).

RPA2 becomes hyperphosphorylated after exposure to IR, UV, certain chemotherapeutic agents, or inhibitors of DNA replication, implicating RPA modification in the cellular responses to DNA damage (9, 10, 11, 12, 13, 14, 15, 16, 17). Identification of the kinases that phosphorylate RPA2 after DNA damage as well as throughout the cell cycle is important for an understanding of the functions and the putative regulatory properties of the protein. Evidence exists that certain cdk-cyclin complexes mediate some of the modifications observed during the progression of cells through the cell cycle (6, 7, 18, 19, 20, 21, 22). Other kinases have been implicated in the phosphorylation of RPA2 after DNA damage.

A kinase with particularly high activity in phosphorylating RPA2 is the DNA-PK (23, 24). Conformational changes occurring on binding of RPA to ssDNA allow a more efficient phosphorylation of RPA2 by DNA-PK (1, 2). In vitro, DNA-PK has been purified as the principal kinase phosphorylating RPA2, and extracts of cells deficient in DNA-PK do not phosphorylate RPA2 (25). RPA-DNA-PK complexes are present in unstressed cells but are disrupted on treatment with camptothecin, an agent that is able to induce DNA DSBs (13, 26). Although these results point to a primary role for DNA-PK in RPA2 phosphorylation, the situation in vivo appears more complex. Thus, irradiation still induces phosphorylation of RPA2 in cells deficient in DNA-PK, although the extent and the kinetics of this phosphorylation are altered (25, 27). These results suggest that a kinase other than DNA-PK also phosphorylates RPA2 in vivo.

RPA2 has been shown to be a phosphorylation target for immunoprecipitates specific for the ATM protein kinase (28), and RPA colocalizes with ATM on synapsed chromosome nodules in mouse cells during the meiotic prophase (29, 30). Furthermore, the IR-induced phosphorylation of RPA2 is delayed in ATM cells (9, 12, 14), indicating that either an ATM-mediated pathway or ATM kinase activity itself plays a role in RPA2 phosphorylation. Consistent with this notion, Mec1, a yeast ATM homologue, is responsible for RPA2 phosphorylation in irradiated Saccharomyces cerevisiae (11). Thus, it is possible that both ATM and DNA-PK contribute to RPA2 phosphorylation, but this possibility has not been studied in detail.

Here we report experiments designed to investigate the role of DNA-PK and ATM in RPA2 phosphorylation after DNA damage. For this purpose we combined genetics with the use of kinase inhibitors. The results indicate a role for both ATM and DNA-PK in the phosphorylation of RPA2 and provide information relevant to the functions of the protein.

HeLa cells were grown in Joklik’s modification of MEM containing 5% iron-supplemented bovine calf serum (Sigma Chemical Co.) and antibiotics. Cells were maintained in the logarithmic phase of growth by subculturing every 4 days at an initial concentration of 10⁶ cells/100-mm tissue culture dish. For experiments, 2 × 10⁶ cells were seeded in 100-mm dishes and allowed to grow for 3 days in a humidified incubator at 37°C, in an atmosphere of 5% CO₂ and 95% air.

M059-J cells (kindly provided by Dr. Joan Allalunis-Turner, University of Alberta, Edmonton, Alberta, Canada) were derived from a human malignant glioma as described previously and found to be deficient in DNA-PKcs (31, 32, 33). They were grown in DMEM supplemented with 10% fetal bovine serum, 1% nonessential amino acids, and 1% l-glutamine, at 37°C in a humidified incubator, in an atmosphere of 5% CO₂ and 95% air. Cells were maintained in a phase of nearly logarithmic growth by subculturing every 4 days at an initial concentration of 0.5 × 10⁶ cells/100-mm tissue culture dish. The same cells were also used to seed cultures for experiments at 0.5 × 10⁶ cells/100-mm dish, which were allowed to grow for 3 days. The growth medium was changed the day before the experiment. At this point, cells reached a density of ∼1.5 × 10⁶/dish and were irradiated to determine RPA2 phosphorylation. Typically, 6–7 × 10⁶ cells were collected per sample. In some experiments we used as a control for M059-J cells M059J/Fus1 cells (kindly provided by Dr. Cordula U. Kirchgessner, University of Stamford, Stamford, CA) grown under similar conditions. These cells have been derived from M059-J by cell fusion with irradiated Scid/hu8 cells containing the human chromosome 8 and retain a fragment of the human chromosome 8 containing the gene for DNA-PKcs (34). M059-J/Fus1 cells express DNA-PKcs and show partial correction for the radiosensitive phenotype of M059-J cells.

AT5BIVA cells (kindly provided by Dr. Colin Arlett, Brunel University, Uxbridge, Middlesex, UK), generated by immortalizing AT5BI, were grown in MEM supplemented with 10% fetal bovine serum and antibiotics. Cells were maintained by routinely subculturing every 4 days at an initial density of 0.5 × 10⁶ cells/100-mm dish. For experiments, cells were seeded at an initial density of 0.5 × 10⁶ cells and were allowed to grow for 3 days at 37°C in a humidified incubator in an atmosphere of 5% CO₂ and 95% air. The growth medium was changed the day before the experiment. All of the chemicals were from Sigma Chemical Co. unless indicated otherwise. UCN-01 was obtained from the drug development branch of the National Cancer Institute.

Cells were irradiated using a Pantak X-ray machine operated at 310 kV, 10 mA with a 2-mm Al filter (effective photon energy ∼90 kV), at a distance of 50 cm and a dose rate of 2.7 Gy/min. Dosimetry was performed with a Victoreen dosimeter that was used to calibrate an in-field ionization monitor.

CEs or NEs were used for the analyses described in this work as indicated in the individual experiments. The characterization of the extract as cytoplasmic or nuclear is based on the procedure of preparation. It does not reflect in a strict sense the localization of proteins in the functioning cell. The procedures used for CE preparation have been described in detail (35, 36) and are presented here only briefly. Cells were trypsinized and collected by centrifugation, washed in PBS and hypotonic buffer, and resuspended for 15 min in three packed cell volumes of hypotonic buffer solution containing 10 mm HEPES (pH 7.8) at 4°C, 1.5 mm MgCl_2, 5 mm KCl, 0.2 mm PMSF, and 0.5 mm DTT. Cells were disrupted by three cycles of freeze (−80°C) and thaw (37°C). Subsequently, 0.11 volumes of high salt buffer containing 10 mm HEPES (pH 7.8) at 4°C, 500 mm KCl (unless indicated otherwise), and 1.5 mm MgCl₂ was added and the homogenate was spun at 3,300 × g for 20 min to precipitate the nuclei. The nuclear pellet is used for the preparation of NE (see below), whereas the supernatant, which is operationally defined as the CE, is collected and spun once again at 14,000 × g for 5 min. When indicated, the resulting CE is dialyzed overnight in 25 mm Tris-HCl (pH 7.5) at 4°C, 10% glycerol, 20% sucrose, 50 mm NaCl, 1 mm EDTA, 0.5 mm DTT, and 0.2 mm PMSF and centrifuged at 14,000 × g for 5 min to remove precipitated protein. Protein concentration is determined using the Bradford assay (Bio-Rad), and aliquots are snap-frozen and stored at −80°C.

To prepare NE, the nuclear pellet obtained after cell disruption and addition of salt is washed with hypotonic buffer containing 50 mm KCl and is resuspended in equal volume (packed nuclear volume) of low salt buffer [20 mm HEPES (pH 7.9) at 4°C, 1.5 mm MgCl₂, 20 mm KCl, 0.2 mm EDTA, 0.2 mm PMSF, and 0.5 mm DTT] containing 800 mm KCl to achieve a final KCl concentration of 400 mm. Nuclei are carefully resuspended and protein extracted for 30 min at 4°C under gentle shaking. After this incubation, extracted nuclei are spun for 15 min at 14,000 × g, the supernatant collected, spun, snap-frozen, and stored in small aliquots at −80°C. Where indicated, the NE is dialyzed for 10 min in low salt buffer containing 20 mm HEPES (pH 7.9) at 4°C, 20% glycerol, 20% sucrose, 100 mm KCl, 0.2 mm EDTA, 0.2 mm PMSF, and 0.5 mm DTT.

Protein (10 μg) from CE and 5 μg from NE are resuspended in 2 × SDS loading buffer [50 mm Tris-HCl (pH 6.8), 2% β-mercaptoethanol, 2% SDS, 0.2% bromphenol blue, and 10% glycerol]. Cell lysate is boiled for 5 min. Proteins are separated by SDS-PAGE on a 12% gel and transferred to polyvinylidene difluoride membrane. Western blot analysis is performed using enhanced chemiluminescence according to the manufacturer (Amersham) and is visualized using the Storm (Molecular Dynamics). RPA antibody (p34–20) was generously provided by Dr. Gilbert Hurwitz, London Health Sciences Centre, Ontario, Canada. Quantitation of Western blots is carried out using the ImageQuant software (Molecular Dynamics) and is shown mainly for the purpose of facilitating comparison between the results obtained after different treatments or with different cell lines.

The SV40-based in vitro DNA replication assay was described previously (37). Briefly, reaction mixtures (25 μl) contain 40 mm HEPES (pH 7.5); 8 mm MgCl₂; 0.5 mm DTT; 3 mm ATP; 200 μm each CTP, GTP, and UTP; 100 μm each of dATP, dGTP, and dTTP; 40 μm dCTP; 40 mm creatine phosphate; 1.25 μg of creatine phosphokinase; 0.15 μg of superhelical plasmid DNA; 100–200 μg of CE; and 0.5 μg of SV40 large T antigen (TAg). The reaction mixture was incubated at 37°C for 1 h. Reactions were terminated by addition of 20 mm EDTA. TAg was prepared as described earlier (37). DNA-PK was purified as described (38).

We investigated IR-induced RPA2 phosphorylation in exponentially growing HeLa cells exposed to 50 Gy of X-rays. Levels and phosphorylation status of RPA2 were separately analyzed in CEs and NEs, because in nonirradiated cells RPA is present in both fractions, and we were interested to investigate whether the fractionation pattern changes in irradiated cells. An additional rationale for this protocol was that the in vitro SV40 DNA replication assay, which we and others use to study the regulation of DNA replication after DNA damage uses exclusively CE (39, 40, 41). Because in such studies RPA surfaces as a candidate regulatory factor (10, 15, 42, 43), it was important to examine its fractionation characteristics after DNA damage.

Extracts were fractionated by SDS-PAGE and RPA2 detected by Western blotting. Fig. 1,A shows the results obtained. In NEs of nonirradiated cells there is a main band at 0 and 24 h and faint bands of slower mobility reflecting phosphorylated RPA2 probably deriving from cells late in the cell cycle (6, 44). Exposure to IR causes strong phosphorylation of RPA2, which in NEs becomes clearly visible at 1 h and peaks at 4 h after irradiation. Phosphorylation is gradually reduced at later times and approaches background 24 h after irradiation (Fig. 1, A and E). A quantitative description of the kinetics of RPA2 phosphorylation is depicted in Fig. 1 E. Plotted is the mean and SE of the relative phosphorylation calculated by image analysis using results from five independent experiments. It is thought that this transient phosphorylation of RPA2 reflects a response of irradiated cells to DNA damage.

Although phosphorylation of RPA2 is clearly observed after irradiation in CE as well (Fig. 1,A), the level is low when compared with the corresponding NE. Thus, phosphorylated RPA2 fractionates preferentially in the nuclear fraction. Similar to observations with NEs, phosphorylation approaches background levels 24 h after irradiation (Fig. 1 A). This reduction in the levels of phosphorylated RPA2 cannot be attributed to selective loss of highly damaged cells, because cell detachment at this time is rather limited.

Extracts analyzed in the above-described experiment were dialyzed in the absence of phosphatase inhibitors before analysis. A comparison between these results and those obtained with nondialyzed samples (last four lanes in Fig. 1 A) indicates that the phosphorylation levels remain unaffected by dialysis. This is relevant for SV40 in vitro DNA replication experiments where extract dialysis is carried out before assembling reactions. Results shown below were obtained using nondialyzed NEs or CEs unless stated otherwise.

The preferential fractionation of phosphorylated RPA2 in the nuclear fraction suggested a tight association with chromatin or the nuclear matrix. To investigate this phenomenon in greater detail, we prepared extracts by gradually increasing the concentration of KCl during the step where the cytoplasmic fraction is generated. Fig. 1 B shows the results obtained. An increase in KCl concentration from 0 to 280 mm increases the amount of phosphorylated RPA2 in the cytoplasmic fraction. These results suggest that phosphorylated RPA associates tightly with nuclear structures and can be extracted only by high salt concentrations.

To examine in greater detail the association of phosphorylated RPA with nuclear structures, we evaluated the level of RPA by Western blotting after extraction at 0.05, 0.4, and 2 m of KCl 4 h after exposure to IR. The results in Fig. 1,C indicate that the majority of phosphorylated RPA2 is released from the nucleus at 0.4 m and that tightly bound RPA2 tends to be in a phosphorylated form after IR. This holds true even for the very small residual fraction remaining on chromatin (“Resid.” in Fig. 1 C) after extraction with 2 m KCl. This finding is in line with earlier observations (45) suggesting that phosphorylated RPA2 binds preferentially to chromatin in nonirradiated cells. In the experiments described below NEs were prepared after release of the cytoplasmic component at 50 mm KCl.

To obtain information on the kinases responsible for RPA2 phosphorylation after IR, we treated cells with wortmannin. Wortmannin inhibits the PI3k at nanomolar concentrations and other kinases of the PI3 family of kinases such as DNA-PK, ATM, and ATR at 1000-fold higher concentrations (46, 47, 48, 49, 50). Because concentrations of wortmannin ≤2 μm did not affect IR-induced RPA2 phosphorylation (results not shown), PI3k can be ruled out as a candidate RPA2 kinase. Fig. 1 D shows results obtained after treatment of HeLa cells with 20 μm of wortmannin given 1 h before irradiation together with the corresponding samples of untreated cells. Wortmannin reduces RPA2 phosphorylation in NEs to background levels for ≤2 h after irradiation. However, at later times RPA2 phosphorylation resumes and reaches a maximum 6 h after irradiation.

Because wortmannin is unstable in aqueous solution, we reasoned that the phosphorylation observed after 2 h may coincide with drug inactivation and the generation of new pools of wortmannin-sensitive kinases. To investigate this possibility we administered 20 μm of wortmannin in three repeats given 1 h before as well as 3 h and 7 h after irradiation (indicated by the arrows in the lower gel of Fig. 1,D). It is evident that repeated treatment with wortmannin reduces to undetectable levels IR-induced RPA2 phosphorylation suggesting that RPA2 phosphorylation depends, directly or indirectly, on wortmannin-sensitive kinases other than PI3k. Because wortmannin binds covalently to the ATP binding site and inactivates irreversibly target kinases, the return of phosphorylation 4 h after a single application indicates not only the resynthesis of the target kinases but also the persistence of the activating signal. The activating signal seems to decay after ∼10 h, because only weak phosphorylation is observed at 24 h in HeLa cells treated with wortmannin 1 h before as well as 3 h and 7 h after irradiation (Fig. 1 D).

From the pool of wortmannin-sensitive kinases, DNA-PK, ATM, and to a lesser degree ATR, are inhibited at the drug concentrations used (50). To obtain additional insight into the actual role of these kinases in RPA2 phosphorylation, we examined other inhibitors with a different spectrum of specificity. Caffeine has been reported recently to inhibit ATM and ATR but not DNA-PK (51, 52, 53, 54). Therefore, we examined the effect of 4 mm of caffeine on RPA2 phosphorylation. The results obtained using NEs are shown in Fig. 2,A together with a set of parallel controls from untreated cells. A quantitative analysis of the results is shown in Fig. 2 C. Caffeine delays and reduces the overall levels of RPA2 phosphorylation implying a role for ATM or ATR in this response. However, clear phosphorylation of RPA2 is still observed suggesting the operation of a caffeine-resistant but wortmannin-sensitive kinase such as DNA-PK. These results and those shown above suggest that more than one kinase is responsible for RPA2 phosphorylation.

Although the above results implicate ATM and ATR in RPA2 phosphorylation, it is not clear whether the effect is direct or indirect. Chk1 is a kinase operating downstream of ATR (55, 56) and is a candidate kinase for the ATR-dependent component of RPA2 phosphorylation. Therefore, we examined the effect of 8-hydroxystaurosporine (UCN-01), an inhibitor of Chk1 (57), on RPA2 phosphorylation. The results in Fig. 2, B and D, show no measurable changes in IR-induced RPA2 phosphorylation after treatment of cells with 1 μm of UCN-01 and suggest that PKc, Chk1, and UCN-01-sensitive cdks (58, 59, 60) are not the primary RPA2 kinases after DNA damage.

The above results in aggregate point to ATM, ATR, and DNA-PK as candidate RPA2 kinases. For additional insight regarding the potential role and the interplay between these kinases in RPA2 phosphorylation we studied M059-J cells. These cells are radiosensitive to killing by IR and have no detectable DNA-PK activity (33, 61). The results obtained are shown in Fig. 3 A. It is notable that no phosphorylation of RPA2 is observed in nonirradiated cells. In NEs, exposure to 50 Gy X-rays induces phosphorylation that is detecTable 4 h after irradiation. The levels of phosphorylation progressively increase at later times, and there is no evidence for decay in phosphorylation even in samples analyzed 24 h after irradiation. On the other hand, results obtained with M059-J/Fus1 cells, generated from M059-J by cell fusion to express DNA-PK (see “Materials and Methods”), show RPA2 phosphorylation patterns similar to those of HeLa cells.

Compared with the results of M059-J/Fus1 cells and those shown in Fig. 1 for HeLa cells, the results with M059-J cells suggest that DNA-PK deficiency causes a delay in RPA2 phosphorylation. Although a reduction in the overall levels of phosphorylation is also occasionally observed, the quantitation shown in Fig. 1,E (includes the results from four experiments with M059-J cells) suggests that these differences are not statistically significant. The response in the CE is qualitatively similar (Fig. 3,A), but as with experiments using HeLa cells, the vast majority of the phosphorylated RPA2 fractionates in the NE. It is notable that in M059-J cells the abundance of phosphorylated RPA2 in the cytoplasmic component is lower than in HeLa cells (see Fig. 1), suggesting a leaky nuclear envelope in the latter cell line, as suggested by other experiments as well (62).

RPA2 phosphorylation is only modestly reduced after a single application of 20 μm of wortmannin in M059-J cells (Fig. 3,A). However, three repeat-applications of 20 μm of wortmannin given according to the protocol outlined in Fig. 1,D additionally reduce RPA2 phosphorylation to nearly undetectable levels (Fig. 3 B). Thus, in line with previous observations (25) and despite the absence of active DNA-PK in M059-J cells, RPA2 phosphorylation is clearly observed.

We probed whether the wortmannin-sensitive kinase operating in M059-J cells is ATM or ATR by treating cells with 4 mm of caffeine. The results obtained are shown in Fig. 3 B together with the corresponding controls of cells incubated in the absence of caffeine. Although the overall levels of RPA2 phosphorylation are lower in this experiment, it is evident that caffeine completely prevents RPA2 phosphorylation.

In contrast to the strong effect of caffeine on IR-induced RPA2 phosphorylation, UCN-01 is without effect (Fig. 4), suggesting that the family of kinases targeted by this inhibitor is not contributing measurably to RPA2 phosphorylation in M059-J cells. This is indicated by the results of the gels shown in Fig. 4,A but also by the quantitative analysis in Fig. 4 B.

Thus, a wortmannin- and caffeine-sensitive kinase phosphorylates RPA2 in M059-J cells in a somewhat delayed but persistent manner. Kinases that satisfy these properties include ATM and ATR.

The above results prompted us to examine the role of ATM in RPA2 phosphorylation. AT cells were exposed to 50 Gy X-rays and RPA2 analyzed in CEs and NEs at different times thereafter. Fig. 5, A and B, show the results obtained with CEs and NEs, respectively. A quantitative analysis of selected results is given in Fig. 5,C as well as in Fig. 1,E. As reported previously by us and others (9, 14, 16), ATM defects lead to delayed phosphorylation of RPA2 compared with normal cells. When results from four experiments are compiled, the kinetics shown in Fig. 1,E are obtained indicating that ATM deficiency results in RPA2 phosphorylation kinetics comparable with that seen in cells deficient in DNA-PK. However, in contrast with the results with DNA-PK-deficient cells, background RPA2 phosphorylation is in AT cells at levels similar to those of wild-type cells. RPA2 phosphorylation is significantly stronger in NE (Fig. 5,B) than in CE (Fig. 5,A) indicating the rather strong association of phosphorylated RPA with nuclear structures in AT cells as well. It is notable that here again RPA2 phosphorylation does not show signs of decay and persists even 24 h after irradiation. This trend is maintained even when the dose of radiation is reduced to 20 Gy (lowest gel in Fig. 5 B).

Repeat-treatment with 20 μm of wortmannin reduces RPA2 phosphorylation to undetectable levels up to 12 h after irradiation. The phosphorylation observed at 24 h probably reflects drug degradation and partial regeneration of the wortmannin-sensitive kinases, and indicates the persistence of the activating signal in AT cells as compared with HeLa cells. Interestingly, in AT cells caffeine at 4 mm is without effect on RPA2 phosphorylation (Fig. 5, B and C). This result suggests that the target of caffeine in this type of experiment is ATM and renders unlikely a contribution by ATR and other caffeine-sensitive kinases. ATR remains active in AT cells. This result is also in line with the lack of RPA2 phosphorylation in vitro by ATR (54) and the observation that inhibition of Chk1, a downstream target of ATR (55), has no effect on RPA2 phosphorylation (see below). The above results suggest that a wortmannin-sensitive but caffeine-resistant kinase phosphorylates RPA2 in AT cells. From the candidate kinases DNA-PK satisfies these requirements.

Fig. 6 shows the effect of UCN-01 on RPA2 phosphorylation in AT cells together with results of parallel samples of untreated cells. It is evident that UCN-01 does not inhibit RPA2 phosphorylation. Quantitative analysis of the results in Fig. 6,A indicates (Fig. 6,B) that in the presence of UCN-01, RPA2 phosphorylation accelerates and reaches kinetics similar to those measured in wild-type cells (compare with the results shown in Fig. 1 E). However, the overall level of phosphorylation remains unaffected.

The above results in aggregate point to ATM and DNA-PK as the RPA2 kinases in vivo and indicate that deficiency in anyone of them leads to delayed phosphorylation that persists longer than in wild-type cells. We were interested to investigate whether the suggested contribution of these two kinases in RPA2 phosphorylation can be reproduced under in vitro DNA replication conditions. Extracts prepared from nonirradiated cells were used to assemble SV40-based DNA replication reactions and the phosphorylation status of RPA2 examined after 1 h of incubation. Fig. 7 shows the results obtained. Low levels of RPA2 phosphorylation are observed in the absence of TAg in extracts of HeLa cells (Fig. 7, Lane 1). Phosphorylation increases significantly in complete reactions (Fig. 7, Lane 2), suggesting the activation of an RPA2 kinase by ongoing DNA replication. This kinase is sensitive to 2 μm of wortmannin as preincubation abolishes RPA2 phosphorylation (Fig. 7, Lane 3). Incubation with 1 mm of caffeine or 1 μm of UCN-01 has no detectable effect on RPA2 phosphorylation (Fig. 7, Lanes 4 and 5) suggesting that, under these in vitro conditions, inhibition of ATM and possibly ATR, as well as Chk1 and possibly other kinases, leaves unchanged RPA2 phosphorylation. Low but detecTable levels of ATM are present in the extract (results not shown).

There is no phosphorylation in reactions assembled with extracts of M059-J cells under any of the conditions examined, suggesting that phosphorylation observed in HeLa cells relies on DNA-PK. This is additionally confirmed by the observation that addition of purified DNA-PK restores RPA2 phosphorylation (Fig. 7, Lanes 10 and 11). However, extracts of M059-J cells effectively support SV40 DNA replication (Ref. 15; results not shown) confirming that RPA2 phosphorylation is not required for DNA replication (8). In AT cells RPA2 phosphorylation is sensitive to wortmannin, and, similar to results obtained with HeLa cells, incubation with caffeine or UCN-01 has no effect on the level of RPA2 phosphorylation.

To mimic the in vivo DNA damage conditions in this in vitro assay, we assembled reactions in which we included 0.1 μg of blunt-end DNA obtained by digesting pUC18 with HaeIII. The presence of DNA in the reactions did not qualitatively alter RPA2 phosphorylation in any of the extracts tested. In summary, the results of the in vitro studies identify DNA-PK as the kinase-phosphorylating RPA2. DNA-PK functions in this system even in the absence of double-stranded DNA. Dependence of RPA2 phosphorylation from the ATM kinase could not be demonstrated under the in vitro conditions used.

The results presented above suggest that RPA2 phosphorylation is mediated through the direct or indirect action of two members of the PI3 family of kinases, DNA-PK and ATM. The lack of caffeine effect on AT cells suggests that ATR is not contributing to a measurable extent to RPA2 phosphorylation. Also, the general inability of UCN-01 to modulate IR-induced RPA2 phosphorylation in all of the cell lines examined here renders unlikely that Chk1 or other kinases targeted by this agent contribute significantly. Because UCN-01 is able to abrogate IR-induced checkpoints including the DNA replication checkpoint (63), the results presented here dissociate RPA phosphorylation from checkpoint activation, in line with earlier observations (12). It is notable that a defect in either ATM or DNA-PK leads to a delay in reaching the maximum level of RPA2 phosphorylation. On the basis of this observation we propose that IR-induced RPA2 phosphorylation is mediated by the coordinated action of ATM and DNA-PK and that both activities are required for rapid phosphorylation. The mechanism underlying the putative coordinated action of ATM and DNA-PK cannot be anticipated on the basis of the results presented here. However, it is possible that the delayed phosphorylation observed after IR in cells defective in either kinase reflects a dampening of the DNA damage-sensing mechanisms and/or a defect in the pathway that removes such damage. Cooperation between ATM and DNA-PK is also inferred by the fact that cells lacking DNA-PKcs have low ATM kinase levels (64) as well as by the observation that DNA ends activate both kinases (24, 65, 66).

The effects noted here with DNA-PKcs-deficient cells are in line with the very low level of phosphorylation observed 2 h after exposure of scid cells to doses in the range between 0 and 20 Gy (27). They are also in line with results obtained with M059-J cells (25) demonstrating a delay in RPA2 phosphorylation and a persistence of this effect up to 6 h after irradiation, the last time point included in the latter report. Differences in RPA2 phosphorylation observed in DNA-PKcs-deficient cells in the above reports may partly reflect differences in the sampling times.

A delay in RPA2 phosphorylation was noted earlier in lymphoblastoid cells derived from AT patients (9) and was confirmed in a wide range of fibroblasts homozygous for the ATM mutation (14); a smaller effect was also observed in fibroblasts heterozygous for the ATM mutation (14). In the latter study, persistence of RPA2 phosphorylation was seen in AT cells exposed to IR. It is interesting that an AT-like phenotype in terms of RPA2 phosphorylation is also generated when wild-type cells are treated with caffeine (Fig. 1), an inhibitor of ATM (52, 53, 54). Furthermore, expression of ATM dominant-negative fragments in normal cells causes a delay in IR-induced phosphorylation of RPA2 (12). Interestingly, however, expression in AT cells of ATM fragments of the kinase domain of the protein do not correct the RPA2 phosphorylation defect, suggesting that functions in addition to kinase activity are required for this end point (12). Such functions may be related to protein-protein or protein-DNA interactions that determine ATM localization and function in the intact cell. Similar functional requirements may lead to the ATM-independent RPA2 phosphorylation under in vitro DNA replication conditions, either in the presence or absence of DNA fragments (Fig. 7), but the possibility that low levels of ATM in the extract prevented manifestation of the effect should be left open. It may be relevant for these observations that ATM associates with chromatin (28), and that the same holds true for a fraction of the cellular RPA (Fig. 1; Ref. 45). Thus, compartmentalization of these proteins may play an important role in their activity and function in the cell. However, it is also possible that ATM-induced RPA2 phosphorylation is indirect and actually mediated by a kinase interacting with ATM in vivo but not in vitro.

Although the persistent phosphorylation of RPA2 after exposure to IR of cells deficient in DNA-PK could be attributed to the persistence of unrejoined DNA DSBs in these cells (61), such an interpretation is not tenable for AT cells, because in these cells, rejoining of DNA DSBs follows nearly normal kinetics (67, 68). Thus, if the presence of phosphorylated RPA2 is the result of persistent signaling from DNA damage, the signal is unlikely to originate from the initially induced DNA DSBs (69, 70). Cells deficient in DNA-PK also show persistent inhibition of DNA replication after exposure to IR, but the relevance, if any, of this observation to the persistence in RPA2 phosphorylation remains unclear (71).

RPA2 can be phosphorylated at several sites localized in the extreme NH₂-terminal region of the protein. Two consensus cyclin-cdk sites are present at Ser-23 and Ser-29 (6), and mutation of these residues to alanine greatly reduces RPA2 phosphorylation in mouse cells (6), in crude extracts of human cells (18), or of recombinant RPA by cyclin B-dependent kinase 1 (cdk1) (20, 22).

For DNA-PK, the primary phosphorylation site has been localized to Thr-21 and Ser-33 (22, 72), but DNA-PK is likely to also phosphorylate other sites in this region of the protein (72). As could be predicted, mutation of Ser-23 and Ser-29 does not affect RPA2 phosphorylation by DNA-PK (18, 20), suggesting that cell cycle and DNA damage-mediated phosphorylation probably serves different functions. Phosphopeptide maps of RPA2 phosphorylated in vivo and in vitro either with cyclin-cdk or with DNA-PK are identical, verifying that the sites determined from in vitro phosphorylation studies are the same as those modified in vivo (6, 72). We assume that RPA2 phosphorylation as observed here reflects modification of Thr-21 and Ser-33, but direct evidence is presently lacking. It is additionally assumed that the sites of RPA2 phosphorylation by DNA-PK and ATM are identical (28). Direct evidence for this has been presented recently (16).

The role of RPA2 phosphorylation in the functions of the holoenzyme, particularly after DNA damage, remains unclear. Some reports are in line with the hypothesis that phosphorylation affects the ability of RPA to support DNA replication. Thus, extracts of human cells exposed to IR or UV, treatments that lead to RPA2 hyperphosphorylation, are deficient in supporting SV40 DNA replication in vitro but can be restored to the activity of nontreated cell extracts by the addition of purified RPA (10, 15). Furthermore, a mutant recombinant RPA lacking the NH₂-terminal 33 residues of RPA2 is inactive in DNA replication in the presence of DNA-PK (73). More recent studies indicate that RPA phosphorylation by DNA-PK or mutation of the RPA2 phosphorylation sites to aspartic acid reduces RPA-TAg complex formation and prevents RPA from physically interacting and stimulating the activity of DNA polymerase α-DNA primase.⁴ Similar effects involving the functional homologues of TAg and pol α-primase in vivo may facilitate a redirection of the RPA activities from DNA replication to DNA repair.

However, other studies question the functional significance of RPA2 phosphorylation. Thus, no significant differences are detected between the activities of wild-type phosphorylated and nonphosphorylated RPA in SV40 DNA replication (8, 74, 75), and deletion of the NH₂-terminal RPA2 residues 2–30 prevents RPA2 phosphorylation but has no effect on the ability of RPA to support in vitro SV40 DNA replication (18, 73). Although the origin of these discrepancies remains unknown, it may well be that they derive from protein functions and interactions taking place in vivo but which are only partially preserved and reproduced in vitro. Indeed, we were unable to reproduce the ATM-dependent RPA2 phosphorylation under in vitro SV40 DNA replication conditions (Fig. 7).

What is the role of RPA2 phosphorylation in DNA metabolism? The results presented here as well as those reported previously (12) dissociate RPA2 phosphorylation from checkpoint activation. Neither the kinetics of RPA2 phosphorylation nor the effects of compounds abrogating checkpoint response (see above) are compatible with such a notion. The fact that RPA2 phosphorylation occurs even in cells with defective checkpoints, such as AT, suggests that it is a late step in the cellular responses to DNA damage.

An attractive hypothesis is that phosphorylation helps to redirect the functions of RPA toward repair. RPA has been implicated directly in HRR (76, 77, 78, 79, 80, 81, 82) and more recently in an in vitro system of NHEJ (83). Because, the fast nature of NHEJ is not in line with the late occurrence of RPA2 phosphorylation, we speculate that RPA phosphorylation signifies an involvement in HRR.

There is ample evidence for the operation of HRR in cells exposed to IR (84, 85, 86, 87, 88, 89). However, recent results from our laboratory using either DT40 mutants with defects in the Rad52 epistasis group of genes (69) or the radiosensitizer caffeine (70) support a model according to which HRR is not removing DSBs from the DNA but functions to restore sequence around a DNA DSB removed, possibly in an error-prone fashion by NHEJ. According to this model HRR activity will reach a maximum a few hours after radiation exposure and after the bulk of IR-induced DNA DSB has been removed by NHEJ. This is in line with the maximum in RPA phosphorylation observed 2–8 h after irradiation in wild-type cells (Fig. 1). According to this model, the delayed phosphorylation in DNA-PK-deficient cells reflects delayed DSB rejoining that impacts on HRR by delaying its inception, thus leading to the persistent RPA2 phosphorylation observed. In the case of ATM, on the other hand, delayed phosphorylation may reflect a defect either in the generation of the HRR sites or a defect in the signaling from them, or both. Indeed, ATM affects homologous recombination (90), and radiation-induced assembly of a Rad51 and Rad52 recombination complex requires ATM (91). An ATM-dependent phosphorylation of RPA2 was also reported in cells exposed to UV light (16). The authors of the latter report also postulate that phosphorylated RPA2 functions in the recombinational postreplication repair of UV lesions.

A different line of investigation also points to a role of RPA in HRR. RPA interacts with p53 (92) through the RPA1 subunit, and this interaction is disrupted after radiation exposure (93). RPA with phosphorylated RPA2 subunit, as present in irradiated cells, fails to interact with p53 (93). Because p53 suppresses homologous recombination in mammalian cells (94), it is possible that RPA2 phosphorylation directs RPA to HRR by abrogating its interaction with p53. Demonstration that phosphorylated RPA functions in HRR will define RPA2 as a marker for this process.

There is evidence in the literature supporting the notion that both ATM and DNA-PK can function as RPA2 kinases (1, 2). The results presented here are also in line with a direct phosphorylation by both kinases. However, recent results for direct phosphorylation in vivo, particularly by ATM, must be considered preliminary at present. Whereas direct phosphorylation is certainly possible, it is also possible that it is initiated by repair processes and carried out by other kinases with sensitivities to caffeine, wortmannin, and UCN-01, as shown here.

We thank Drs. J. Allalunis-Turner and C. Arlett for cells, and Dr. G. Hurwitz for antibodies. We also thank Nancy Mott for secretarial help and for editing the manuscript.